[Bugfix] Fix tool_calls Iterable consumed when debug logging is enabled#34844

Merged

vllm-bot merged 103 commits intovllm-project:mainfrom

wojciech-wais:fix/tool-calls-iterable-consumed-debug-logging

Apr 15, 2026

Commits on Mar 18, 2026

[Bugfix] Fix tool_calls Iterable consumed when debug logging is enabled
wojciech-wais
committed

Commits on Apr 1, 2026

[Bugfix] Switch tool_calls materializer to mode=before to catch generators
wojciech-wais
committed
Merge branch 'main' into fix/tool-calls-iterable-consumed-debug-logging
wojciech-wais
authored

Commits on Apr 2, 2026

[Bugfix] Use both before+after validators to fully fix tool_calls
wojciech-wais
committed

Commits on Apr 9, 2026

Merge branch 'main' into fix/tool-calls-iterable-consumed-debug-logging
mergify[bot]
authored

Commits on Apr 13, 2026

[Refactor] Move NVFP4 GEMM management into NvFp4LinearKernel (#39129)

mgoin
authored and
wojciech-wais
committed
[XPU] check is_xccl_available before oneccl warmup (#39302)

xinyu-intel
authored and
wojciech-wais
committed
[ASR] Fix spacing bw chunks in multi chunk audio transcription (#39116)

ekagra-ranjan
authored and
wojciech-wais
committed
Add nightly b200 test for spec decode eagle correctness (#38577)

puririshi98
authored and
wojciech-wais
committed
[Spec Decode] fix returning size mismatch on extract hidden states proposer (#38610)

authored and
wojciech-wais
committed
[CI/Build[ Don't auto-rebase PRs with CI failures (#39443)

DarkLight1337
authored and
wojciech-wais
committed
[LMCache] vLLM Block Allocation Event (#38856)

authored and
wojciech-wais
committed
[Model][Perf] Enable checkpoints prefetching for Lustre FS by default (#39422)

arpera
authored and
wojciech-wais
committed
fix(gdn): Align prefill warmup with real prefill path (#39169)

ibrahim1023
authored and
wojciech-wais
committed
[ROCm] Fix AITER ops fake impl and minor bugs (#36092)

ChuanLi1101
authored and
wojciech-wais
committed
[Feature] Add auto-detection for reasoning_config when only reasoning_parser is set (#38214)

chaunceyjiang
authored and
wojciech-wais
committed
[CT][FP8][Marlin] refactor CompressedTensorsW8A16Fp8 to use kernel abstraction (#38244)

jikunshang
authored and
wojciech-wais
committed
feat(cpu): add CPU support for draft model speculative decoding (#32662)

ganeshr10
authored and
wojciech-wais
committed
[BugFix] Fix OOB read in CUTLASS grouped GEMM with epilogue (#38571)

authored and
wojciech-wais
committed
Add EXAONE-4.5 (#39388)

authored and
wojciech-wais
committed
[Mergify] Update model vendor auto-label rules (#39312)

DarkLight1337
authored and
wojciech-wais
committed
[KV Offload] Implement `shutdown()` in `OffloadingConnector` and related classes (#39182)

ronensc
authored and
wojciech-wais
committed
[feat]: make DCP error msg clearer (#28443)

authored and
wojciech-wais
committed
[Bugfix] Fix broken explicit unquantized kv cache dtype support (#38922)

Isotr0py
authored and
wojciech-wais
committed
[BugFix][CPU] Add CPU profiler summary file output (#38366)

Elm8116
authored and
wojciech-wais
committed
Add platform manual_seed_all API (#38468)

yma11
authored and
wojciech-wais
committed
[Core] Change max_model_len in EngineCoreReadyResponse to be non-None (#39442)

njhill
authored and
wojciech-wais
committed
[Perf] Reduce H2D pageable memory copies (#38794)

authored and
wojciech-wais
committed
update CODEOWNERS file (#39439)

xuechendi
authored and
wojciech-wais
committed
[GGUF] Support non-standard quant types with prefix (e.g. UD-IQ1_S) (#39471)

authored and
wojciech-wais
committed
[compile] Allow strings in custom ops without regressing compilation times (#38123)

zou3519
authored and
wojciech-wais
committed
Nemotron Nano VL: Streamline pixel shuffle (#37580)

milesial
authored and
wojciech-wais
committed
[Bugfix] Fix V1 dummy run writing NaN to KV cache null block (#39444)

authored and
wojciech-wais
committed
[PluggableLayer][3/N] Apply PluggableLayer to llm_head and vocab embedding layer (#33465)

authored and
wojciech-wais
committed
[ZenCPU] Make PT Backport Patch Accessible to vLLM (#38205)

authored and
wojciech-wais
committed
[model] support FireRedLID (#39290)

authored and
wojciech-wais
committed
[Kernel][Hardware][AMD] Add TritonW4A16LinearKernel for ROCm (#37352)

authored and
wojciech-wais
committed
[Performance] Remove unnecessary zero-fill of MLA decode output tensor in Aiter backend (#37539)

xaguilar-amd
authored and
wojciech-wais
committed
[ROCm] Add RDNA 3.5/4 device IDs (gfx1150, gfx1151, gfx1201) (#38455)

authored and
wojciech-wais
committed
[CI] Add Nixl+OffloadingConnector e2e integration tests (#39200)

NickLucche
authored and
wojciech-wais
committed
[Model] Implement LoRA support for Qwen3ASRForConditionalGeneration (#37247)

petern48
authored and
wojciech-wais
committed
[New Model]: jinaai/jina-reranker-v3 (#38800)

noooop
authored and
wojciech-wais
committed
[Model Runner V2] support auto resolve cudagraph mode/sizes based on attn backend (#32936)

izhuhaoran
authored and
wojciech-wais
committed
[Kernel] Porting the TRTLLM minimax_allreduce_rms kernels (#37045)

jeejeelee
authored and
wojciech-wais
committed
[ROCm] [AITER] Revert AITER version to v0.1.10.post3 (#39509)

tjtanaa
authored and
wojciech-wais
committed
feat: add logit_scale to PoolerConfig for affine score calibration (#39435)

authored and
wojciech-wais
committed
perf(moe): add tuned fused_moe config for RTX PRO 6000 Blackwell Server Edition (#39183)

efortin
authored and
wojciech-wais
committed
[CI] Add MultiConnector (Nixl+Offloading) e2e edge case tests (#39343)

ZhanqiuHu
authored and
wojciech-wais
committed
[Bugfix] Fix FlashInfer crash with kv_cache_dtype_skip_layers (#39002)

yzong-rh
authored and
wojciech-wais
committed
Add Gemma4 Eagle3 support (#39450)

authored and
wojciech-wais
committed
Fix pre-commit labeled trigger system (#39523)

fynnsu
authored and
wojciech-wais
committed
Add structure to `requirements/` directory (#39024)

authored and
wojciech-wais
committed
[Refactor] Move MXFP8 GEMM management into MxFp8LinearKernel (#39205)

mgoin
authored and
wojciech-wais
committed
[Docs] Use `--torch-backend=auto` for editable install docs (#39511)

mgoin
authored and
wojciech-wais
committed
Revert "Add nightly b200 test for spec decode eagle correctness (#38577)" (#39512)

benchislett
authored and
wojciech-wais
committed
[XPU] Fix spec-decode UTs under tests/v1/spec_decode (#38491)

authored and
wojciech-wais
committed
Fix the order of _free_encoder_inputs (#38907)

authored and
wojciech-wais
committed
[Bugfix] add SupportsMultiModal to Exaone4_5_MTP (#39526)

elwhyjay
authored and
wojciech-wais
committed
[Misc] Update deprecation warning for --model flag (#39518)

z1ying
authored and
wojciech-wais
committed
[CI/Build] Fix sentence-transformers version in CPU test (#39557)

bigPYJ1151
authored and
wojciech-wais
committed
[Bugfix] Fix GDN FLA kernel crashes with NULL_BLOCK_ID=0 CUDA graph padding (#39064)

authored and
wojciech-wais
committed
[Gemma4][Bugfix]: Enable Gemma4ForCasualLM to load lora adapters correctly (#38844)

ShubyM
authored and
wojciech-wais
committed
[Perf] Fuse Zero Initializer for FP8 DeepGemm Block Quant Kernel (#39547)

wzhao18
authored and
wojciech-wais
committed
[Bugfix] Runtime driver check for cuMemcpyBatchAsync in swap_blocks_batch (#38919)

authored and
wojciech-wais
committed
[Quant] add CompressedTensorsW8A8Mxfp8 for linear and MoE layers (#38815)

EdalatiAli
authored and
wojciech-wais
committed
[XPU][CT] support per-channel quantization in xpu fp8 linear method (#38316)

yma11
authored and
wojciech-wais
committed
[ROCm][CI/Build] Fix memory cleanup in MM test (#39555)

AndreasKaratzas
authored and
wojciech-wais
committed
Support FP8 KVCache on XPU (#37731)

authored and
wojciech-wais
committed
fix(kimi_k25): resolve media_placeholder_token_id from tokenizer (#39344)

authored and
wojciech-wais
committed
[HMA] [KVEvent] Enable GPU-side KV events for HMA (#37688)

authored and
wojciech-wais
committed
[Pooling] Disable async scheduling by default for pooling models (#39592)

njhill
authored and
wojciech-wais
committed
[Core][Metrics] Remove `vllm:prompt_tokens_recomputed` metric (#38709)

markmc
authored and
wojciech-wais
committed
[KVConnector][NIXL] Organize NIXL connector into its own directory (#39354)

NickLucche
authored and
wojciech-wais
committed
fix(moe): fix RoutedExpertsCapturer assertion failure with DP>1 and MK path (#37879)

Young-Leo
authored and
wojciech-wais
committed
[compile] Bug fix for _decompose_size_nodes (#38360)

anijain2305
authored and
wojciech-wais
committed
[Bugfix] [Tests] Enforce `out` tensor device in `kernel/moe/test_cutedsl_moe.py` (#39644)

zyongye
authored and
wojciech-wais
committed
[ROCm][CI] Removed stale tests and extended acceptance test (#39651)

AndreasKaratzas
authored and
wojciech-wais
committed
[XPU] revert torch-xpu to 2.10 (#39656)

jikunshang
authored and
wojciech-wais
committed
[LoRA] Support dual CUDA streams-Linear Layer (#35721)

jeejeelee
authored and
wojciech-wais
committed
[compile] Enable AOT compile with batch invariance mode. (#39201)

authored and
wojciech-wais
committed
fused qknorm+rope kernel optimization for SM9.0 (#37376)

authored and
wojciech-wais
committed
fix(lmcache): correct store for cached requests and num_scheduled_tokens in lmcache_mp_connector.py (#39655)

maobaolong
authored and
wojciech-wais
committed
feat: rename logit_bias/logit_scale to logit_mean/logit_sigma for affine score calibration (#39530)

authored and
wojciech-wais
committed
[XPU]Enhance environment collection for Intel XPU and optimize layout (#35698)

authored and
wojciech-wais
committed
[Bugfix] Fix GLM tool parser streaming with MTP or stream interval (#39253)

sfeng33
authored and
wojciech-wais
committed
feat: add max_tokens_per_doc in rerank request. (#38827)

authored and
wojciech-wais
committed
[Bugfix] Fix Responses API instructions leaking through previous_response_id (#37727)

authored and
wojciech-wais
committed
[MXFP8] [XPU] add a new compressed tensor schema and add a xpu mxfp8 gemm kernel (#38707)

zufangzhu
authored and
wojciech-wais
committed
[Bug] Fix TypeError when hf_config.architectures is None during model loading (#38849)

TihoElek
authored and
wojciech-wais
committed
[Bugfix] stream failure when model name not in audio endpoints (#36679)

ekagra-ranjan
authored and
wojciech-wais
committed
[Bugfix][CT] Fix KV cache scale handling (#39418)

yiliu30
authored and
wojciech-wais
committed
[Bug] Fix rocm sparse attn indexer issue (#39225)

authored and
wojciech-wais
committed
[CI][Metrics] Fix local_cache_hit assertion after prompt tokens metrics updates (#39709)

ZhanqiuHu
authored and
wojciech-wais
committed
[Bugfix] Fix tensor shape mismatch in sparse attention with speculative decoding (#39542)

santiramos27
authored and
wojciech-wais
committed
[Quantization] [Refactor] Create special "GptOssMxfp4MoeMethod" (#39604)

zyongye
authored and
wojciech-wais
committed
[Bugfix] Use is_integrated to detect UMA GPUs for memory reporting (#35356)

authored and
wojciech-wais
committed
[Misc] Multi-turn benchmark output performance json (#39572)

NickLucche
authored and
wojciech-wais
committed
[Bugfix][Kernel][ROCm] Fix triton_w4a16 scales mismatch when BLOCK_K > group_size (#39705)

JartX
authored and
wojciech-wais
committed
[Doc] Fix Python-only build 404 fallback guidance (#38052)

authored and
wojciech-wais
committed
[Bugfix] Reject non-nvfp4 dtypes when using the flashinfer_nvlink_one_sided all2all backend (#39717)

authored and
wojciech-wais
committed
[Core][Metrics] expose waiting request breakdown via labeled metric (capacity/deferred) (#38435)

authored and
wojciech-wais
committed
[Bugfix][Pooling] Fix silent weight corruption with buffer-reusing iterators (#39650)

pedramr
authored and
wojciech-wais
committed
Merge branch 'main' into fix/tool-calls-iterable-consumed-debug-logging
wojciech-wais
authored